arX

iv:1

301.

0622

v1 [

astr

o-ph

.CO

] 3

Jan

201

3

[O iii] Emission and Gas Kinematics in a Lyman-alpha Blob at z ∼

3.1

Emily M. McLinden1, Sangeeta Malhotra2, James E. Rhoads2, Pascale Hibon3, Anne-Marie

Weijmans4, Vithal Tilvi5

ABSTRACT

We present spectroscopic measurements of the [O iii] emission line from two

subregions of strong Lyα emission in a radio-quiet Lyman-alpha blob (LAB).

The blob under study is LAB1 (Steidel et al. 2000) at z ∼ 3.1, and the [O iii]

detections are from the two Lyman break galaxies embedded in the blob halo.

The [O iii] measurements were made with LUCIFER on the 8.4m Large Binocular

Telescope and NIRSPEC on 10m Keck Telescope. Comparing the redshift of

the [O iii] measurements to Lyα redshifts from SAURON (Weijmans et al. 2010)

allows us to take a step towards understanding the kinematics of the gas in the

blob. Using both LUCIFER and NIRSPEC we find velocity offsets between the

[O iii] and Lyα redshifts that are modestly negative or consistent with 0 km

s−1 in both subregions studied (ranging from -72 ± 42 – +6 ± 33 km s−1).

A negative offset means Lyα is blueshifted with respect to [O iii] a positive

offset then implies Lyα is redshifted with respect to [O iii]. These results may

imply that outflows are not primarily responsible for Lyman alpha escape in this

LAB, since outflows are generally expected to produce a positive velocity offset

(McLinden et al. 2011). In addition, we present an [O iii] line flux upper limit

on a third region of LAB1, a region that is unassociated with any underlying

galaxy. We find that the [O iii] upper limit from the galaxy-unassociated region

of the blob is at least 1.4 – 2.5 times fainter than the [O iii] flux from one of the

LBG-associated regions and has an [O iii] to Lyα ratio measured at least 1.9 –

3.4 times smaller than the same ratio measured from one of the LBGs.

1McDonald Observatory, University of Texas at Austin, Austin, TX 78712, USA

2School of Earth and Space Exploration, Arizona State University, Tempe, AZ 85287

3Gemini Observatory, La Serena, Chile

4Dunlap Institute for Astronomy & Astrophysics, University of Toronto, 50 St. George Street, Toronto,

ON M5S 3H4, Canada

5George P. and Cynthia Woods Mitchell Institute for Fundamental Physics and Astronomy, and Depart-

ment of Physics and Astronomy, Texas A&M University, College Station, TX, 77843

– 2 –

Subject headings: galaxies: high redshift — intergalactic medium — galaxies:

kinematics and dynamics

1. INTRODUCTION

Lyman-alpha (Lyα) first became a useful tool for observing high-z sources with the

discovery of large samples of Lyα emitting galaxies (LAEs) (Cowie & Hu 1998; Hu et al.

1998; Rhoads 2000). The same narrowband imaging techniques that uncovered LAEs be-

gan uncovering a different set of objects that were also very bright in Lyα. These rarer,

more extended, and more luminous objects are what we now call Lyα blobs (LABs) (e.g.,

Steidel et al. 2000; Francis et al. 2001; Matsuda et al. 2004; Dey et al. 2005; Nilsson et al.

2006; Prescott et al. 2012). LABs are extremely large (∼ 30–200 kpc) radio-quiet Lyα neb-

ulae in the high redshift universe. LABs are highly luminous (LLyα ∼ 1043−44 ergs s−1), and

yet despite rigorous study in the last decade, the mechanism(s) that power this immense Lyα

flux is not fully understood. This paper will focus on investigating the kinematics of and

mechanisms powering such objects by investigating LAB1, a z ∼ 3.1 LAB first discovered

by Steidel et al. (2000).

There are currently three most widely discussed scenarios to explain both the large

spatial extent and powerful Lyα flux of these blobs. The first of these is that the gas in

LABs is heated by photoionization from massive stars and/or AGN (Geach et al. 2009).

A second scenario proposes that gas in LABs is heated by cooling flows / cold accretion

(Haiman et al. 2000; Dijkstra & Loeb 2009). Finally some authors have proposed LABs

originate from overlapping supernova remnants from massive stars after a powerful starburst

(Taniguchi & Shioya 2000; Ohyama et al. 2003) producing superwinds.

Adding to the controversy, observations in recent years from different authors have led

to different conclusions about which of these scenarios are responsible for said observations.

Nilsson et al. (2006) have argued that their observations of a z∼ 3.16 LAB were best matched

by cooling flows onto a dark matter halo. This is in contrast to the conclusions of Hayes et al.

(2011), who found evidence of polarized Lyα radiation in LAB1. The Hayes et al. (2011)

results suggest that Lyα photons are scattered at large radii from their production sites and

this observation seems to not only favor the role of scattering in outflows in LABs, but the

authors contend their discovery can actually rule out most inflow models. But a similar study

by Prescott et al. (2011) found no evidence of polarization in a LAB at z ∼ 2.656 and these

authors argue their results are inconsistent with spherical outflows and Lyα scattering from

nearby AGN. Yet another conclusion is reached by Yang et al. (2011) whose observations of

Lyα and Hα emission in two z ∼ 2.3 LABs rule out simple infall models and models that rely

– 3 –

on large outflows. This diversity of conclusions may mean that there are diverse mechanisms

powering different blobs (or multiple mechanisms at play in single blobs) or it may mean

that truly conclusive observations have not yet been presented.

To try to provide new data to differentiate amongst the possible LAB sources we focus

on LAB1 (Steidel et al. 2000). LAB1 resides in SSA22, a protocluster region at z ∼ 3.1

(Steidel et al. 1998). LAB1 extends ∼ 100 kpc (Weijmans et al. 2010) and has a Lyα lumi-

nosity of 1.1 × 1044 erg s−1 (Matsuda et al. 2004). This makes LAB1 one of the brightest

and most spatially extended LABs yet observed (Steidel et al. 2000; Matsuda et al. 2004).

LAB1 is comprised of five separate regions of Lyα emission known as C11, C15, R1, R2, and

R3 (see Figure 1 of Weijmans et al (2010)). C11 and C15 are both Lyman break galaxies

(LBGs) identified by Steidel et al. (2000). R3 has been identified as an extremely red galaxy

(Geach et al. 2007) and may or may not be associated with a bright submillimeter source

and nearby radio source (Chapman et al. 2001, Chapman et al. 2004, but see also Kohno

et al. 2008, Yang et al, 2012). R1 and R2 are not identified with galaxies (Weijmans et al.

2010). Geach et al. (2009) also demonstrated that AGN activity is not significant in LAB1

as LAB1 is not detected in a 400 ks Chandra exposure. We note that Matsuda et al. (2004)

identified C15 as a separate LAB, but in this paper we refer to all five subregions as com-

ponents of LAB1, following the nomenclature of Steidel et al. (2000) and Weijmans et al.

(2010).

In this paper we present new spectroscopic [O iii] observations of LAB1. We focus on

the two Lyα subregions C15 and C11, the two parts of the larger LAB1 structure in which

we detected [O iii]. The use of [O iii] to study the kinematics of LAB1 is powerful because

[O iii] is not subject to resonant scattering as Lyα is. Comparing [O iii] to Lyα emission

allows us to characterize any systemic offsets between Lyα and [O iii], and thereby detect

the presence of outflows or inflows. We first demonstrated the efficacy of this method in a

sample of LAEs in McLinden et al. (2011).

Our new [O iii] data presented here is compared to Lyα data originally presented in

Weijmans et al. (2010). Weijmans et al. (2010) measured Lyα from each of the 5 subregions

in LAB1 with the integral field spectrograph SAURON over 23.5 hours (including 9 hours

of SAURON data originally obtained by Bower et al. (2004)). We obtained the reduced

datacube produced from these observations from Weijmans. This allows us to extract Lyα

profiles at locations corresponding to our NIR [O iii] observations for careful comparison of

[O iii] and Lyα redshifts. Henceforth we refer to Weijmans et al. (2010) as W10.

In Sections 2 and 3 we present our [O iii] detections from two near-infrared (NIR)

spectrographs (NIRSPEC and LUCIFER). In Section 5 we look for any velocity offsets

between our measured [O iii] redshifts and those of Lyα to look for any evidence of inflows

– 4 –

or outflows and in Section 6 we discuss the implications of our findings of velocity offsets

(∆v) that are modestly negative or consistent with zero. We also compare our results to

other authors and explore if there are currently any Lyα radiative transfer models that can

match our results. Where relevant, we adopt the standard cosmological parameters H0 = 70

km s−1 Mpc−1, Ωm = 0.3, and ΩΛ = 0.7 (Spergel et al. 2007). We use the following vacuum

wavelengths, 1215.67 A for Lyα, 3727.092/3729.875 A for [O ii], 4862.683 A for Hβ, and

4960.295/5008.240 for [O iii] from the Atomic Line List v2.041.

2. NIRSPEC DATA AND REDUCTION

We initially detected [O iii] emission from LAB1 using the near-infrared spectrograph

NIRSPEC (McLean et al. 1998) on the 10m Keck II telescope on 6 August 2010 (UT). We

used the low-resolution mode of NIRSPEC, with the 42 × 0.76′′ slit and the NIRSPEC-6

filter which covers 1.558 – 2.315 µm. The spectral resolution of NIRSPEC in this setup is

∼ 290 km s−1. This filter encompasses the redshifted (z ∼ 3.1) [O iii] doublet and covers

redshifted Hβ as well, though we did not see this line. We completed three 500-second

integrations, for a total exposure of 25 minutes. The longslit was oriented so that LAB1

regions C15 and C11 (Steidel et al. 2000) both lie directly on the slit. Region R2 also has

some peripheral coverage, though not directly through the location of its peak Lyα emission.

Due to the short length of slit we were unable to place an additional continuum-bright object

on the slit, so LAB1 was acquired via blind-acquisition from a nearby star.

[O iii] detections from C15 and C11 are evident in single, raw 500-second exposures,

when a second frame is subtracted from the frame of interest. We find these detections at

their expected locations in the spatial direction along the slit, and find they also have the

separation from one another that we expect for C11 and C15. In addition, the detections are

in the wavelength range expected of each region’s Lyα redshift. The dither pattern we used

is also clearly visible in positive and negative detections when we perform this sort of quick-

look subtraction, providing assurance that these detections are not transient cosmic rays in

a single exposure. These facts combined give us confidence that the emission we detect is,

in fact, from [O iii] emission from C15 and C11. See the top panel of Figure 2 for these

detections in a sky-subtracted 2D frame. The [O iii] emission from C15 is strong and comes

through as such in both our 2D and 1D reduction processes detailed below. The [O iii]

detections from C11 appear much fainter and are not as evident, though still marginally

detected, throughout our 2D and 1D reduction processes.

1http://www.pa.uky.edu/∼peter/atomic/index.html

– 5 –

Initial data reduction of each 500-second exposure was done using NIRSPEC REDUCE,

a set of IDL programs written by G. D. Becker specifically for reduction of NIRSPEC longslit

data. We used these scripts for flat fielding and sky subtraction in each exposure. The

sky subtraction process in NIRSPEC REDUCE is based on the algorithm of D. Kelson

(Kelson 2003) which provides excellent sky subtraction of even tilted skylines, such as those

in NIRSPEC longslit data. The outcome of these reduction steps are three individual,

2D, flat-fielded, sky-subtracted exposures. We also reduced an argon lamp exposure and a

standard star exposure in this same way.

We then fed these exposures into IRAF procedures in the WMKONSPEC package2.

Each frame was rectified to a horizontal-vertical grid in x- y using the tasks XDISTCOR

AND YDISTCOR, which remove x and y distortion in the images, respectively. Once the

exposures were rectified, we used IMCOMBINE to median combine the frames into a single

exposure. We specified offsets in the IMCOMBINE procedure to remove dithers along the

slit that were performed during our observations. The result, what we call our reduced 2D-

spectrum, is shown in Figure 2. The locations of our [O iii] detections in C11 and C15 are

circled.

To extract 1D spectra we used the IRAF DOSLIT procedure. We first defined an

aperture trace using a bright standard star observation which can be easily traced along the

entire slit. We then transferred this aperture to the correct spatial locations in our science

exposure to extract spectra of C15 and C11. We used this transferred aperture since neither

region has continuum emission that we are able to trace for aperture creation. The spectra

were wavelength calibrated using an argon lamp. The RMS error from wavelength calibration

was 0.67 and 0.45 A for C15 and C11 respectively. We used the IRAF WMKONSPEC task

SKYINTERP to remove remaining residuals from sky lines. The resulting 1D spectra from

C15 and C11 are shown in Figure 2.

We flux calibrate our NIRSPEC spectra using a magnitude 8.85 (V band) A1V star

that was observed in the same setup as our science observations (0.76′′ slit). We averaged an

A0V and an A2V Pickles model spectrum to approximate an A1V spectrum. We then scaled

down the A1V stellar spectrum (Pickles 1998) to match the magnitude of the observed star.

We created a sensitivity function with units of erg cm−2 A−1 counts−1 by dividing the model

spectrum by the observed stellar spectrum and multiplying by the length of the observation.

This sensitivity function is then multiplied by the extracted NIRSPEC spectra for C15 and

C11, and then the result is divided by the integration time for each object to produce flux

calibrated spectra in units of erg s−1 cm−2 A−1.

2http://www2.keck.hawaii.edu/inst/nirspec/wmkonspec.html

– 6 –

Fig. 1.— Orientation of longslit for NIRSPEC and LUCIFER observations. NIRSPEC slit

is in white, LUCIFER slit is magenta. Boxes indicate locations of subregions C11,C15, and

R1-R3, from W10. Image from STIS-MIRVIS observations, retrieved from HST archives

(Chapman et al. 2003).

– 7 –

Fig. 2.— Top image is median combined, sky-subtracted, distortion corrected 2D spectrum

from NIRSPEC. Top image has been smoothed with a Gaussian function with radius of

three pixels. Emission lines from C11 and C15 are circled. [O iii] doublet from C15 are

upper circles, 5008.24 A line from C11 is lower circle. Bottom row contains 1D NIRSPEC

spectra for C15 and C11 (top panel) and extracted background. Best-fit Gaussians are

overlaid on emission lines - see Section 5.1. Blue is 4960.295 A line and red is 5008.24 A line.

Extracted background has been multiplied by -1 so one can easily recognize the sky lines,

but the sky lines were actually negative in some places due to imperfect sky subtraction in

the 2D image.

– 8 –

3. LUCIFER DATA AND REDUCTION

We subsequently made additional NIR observations of LAB1 using LUCIFER (LBT

NIR Spectrograph Utility with Camera and Integral-Field Unit for Extragalactic Research)

on the 8.4m LBT (Seifert et al. 2003; Ageorges et al. 2010). We used the longslit mode of

LUCIFER with a 1′′ slit utilizing the H+K grating with 200 lines/mm and the N1.8 camera.

LUCIFER has a spectral resolution of ∼ 318 km s−1 near 1.6 µm and ∼ 233 km s−1 near

2.2 µm. We combine three 300-second integrations, for a total exposure of 15 minutes. We

placed the longslit at a slightly different orientation than our NIRSPEC observations in

hopes of capturing more emission from R2 and C15. For the LUCIFER setup, C15 and R2

lay directly on the slit, with part of C11 on the edge of the slit (see Figure 1). Given the

length of the LUCIFER longslit, we were able to place a galaxy with continuum on the slit

Fig. 3.— Third NIRSPEC frame subtracted from second NIRSPEC frame, before x-axis and

y-axis distortion correction, and sky subtraction. [O iii] emission (5008.24 A line) from C11

and C15 is more evident than in Figure 2. Emission from C15 in cyan circles, emission from

C11 in green circles. Emission from second frame is black (positive), emission from third

frame is white (negative). Positive-negative dither pattern is clear, showing a detection in

both frames displayed here.

– 9 –

as well. This aids in aperture extraction during the reduction process.

We reduced the LUCIFER data in a very similar manner as the NIRSPEC data, but we

used a modified version of the NIRSPEC REDUCE package to accommodate the different

detector size and orientation of the LUCIFER data. After the NIRSPEC REDUCE proce-

dures, the individual exposures were again median combined with IMCOMBINE and offsets

from dithering along the slit were accounted for. Figure 4 shows the combined 2D spectrum

after this step. For the 1D extraction, we created an aperture trace using the continuum

source that shared the slit with our science targets, instead of the standard star as in our

NIRSPEC procedure. Then we shifted the aperture to the correct spatial locations to ex-

tract 1D spectra for C15, C11 and R2. The spectra were again wavelength calibrated with

an argon lamp exposure. Figure 4 shows the 1D extraction of C15 from our LUCIFER data,

which was the only region from which we detected [O iii] in our LUCIFER observations. We

flux calibrate our LUCIFER spectra using a magnitude 6.16 (V band) A5V star that was

observed in the same setup as our science observations (1′′ slit). We used a Pickles A5V

model spectrum and otherwise the calibration process is otherwise the same as described in

Section 2.

Fig. 4.— Left plot is 2D LUCIFER spectrum of C15. The image has been smoothed by

a Gaussian with a radius of 3 pixels and the [O iii] detection (5008.24 A) is circled. Right

top panel is extracted 1D LUCIFER spectrum of C15. Feature at ∼ 20355 A is a bad

column. Best-fit Gaussian is overlaid on emission line - see Section 5.1. Bottom right panel

is extracted background - only minimal sky line residuals remain because sky subtraction in

2D spectrum was excellent. Extracted background had been multiplied by -1.

– 10 –

4. Lyα DATA

We extracted Lyα profiles from the reduced SAURON datacube, data that was initially

presented in W10. The SAURON data was registered to the R band image that was used to

align our NIRSPEC and LUCIFER slits. Next, we laid our LUCIFER and NIRSPEC slits

(Figure 1) on the registered SAURON data and extracted the spaxels in the SAURON data

that corresponded with the LUCIFER and NIRSPEC [O iii] detections. This is essentially

creating a virtual longslit aperture for the SAURON data. The LUCIFER longslit was 1′′

wide and the NIRSPEC longslit was 0.76′′ wide, so our virtual longslit for the SAURON

data was 3 spaxels across, or 1.2′′ to provide an approximate match to the LUCIFER and

NIRSPEC apertures. The extracted spaxels were summed to create 1D spectra for C11 and

C15. Because we observed C15 with both NIRSPEC and LUCIFER, with slightly different

slit orientations, we extracted two spectra for C15 - one that corresponds to our NIRSPEC

slit orientation, and one that corresponds to our LUCIFER slit orientation. When discussing

results for C15 below we specify whether the results are for C15 from the NIRSPEC alignment

or C15 from the LUCIFER alignment. C11 was only detected with NIRSPEC, so we only

extracted one spectrum for C11. The Lyα lines detected in the 1D spectra described above

are shown in Figure 5.

To get error bars for the flux in the extracted 1D spectra we selected 1000 random spaxels

in the SAURON datacube that were outside the area of the blob. At each wavelength we

created a histogram of the 1000 flux values found in our random spaxels at that wavelength,

fit a Gaussian to that histogram, and adopted the sigma of that Gaussian as the 1 sigma

error on the flux at that wavelength.

5. RESULTS

5.1. [O iii] Redshifts

As we did in McLinden et al. (2011) we fit detected [O iii] lines with a single symmetric

Gaussian plus a constant, using the IDL routine MPFITEXPR3. We fit the NIRSPEC and

LUCIFER spectra independently. The central wavelength of the best fit Gaussian deter-

mines the [O iii] redshift. We fit the 4960.295 A and 5008.24 A lines independently for C15

in NIRSPEC and we find only the 5008.24 A line in the LUCIFER data for this galaxy.

Given the redshift of this region from the 5008.24 A line in LUCIFER, the 4960.295 A line

3http://cow.physics.wisc.edu/ craigm/idl/down/mpfitexpr.pro

– 11 –

should fall at ∼ 20332.3 A, right on the edge of the 20339.497 A (vacuum) OH emission line

(Rousselot et al. 1999). This may explain why, after sky interpolation we are unable to detect

this line in the slightly lower resolution of LUCIFER. The agreement between the redshift

derived from 4960.295 and 5008.24 A lines in NIRSPEC spectrum is good (see Table 1). We

take the average of the 4960.295 A and 5008.24 A redshift as the derived systemic redshift

for C15 from NIRSPEC, and use the redshift of the single line for C15 from LUCIFER.

Only the stronger 5008.24 A line was detected in C11, and only in the NIRSPEC

spectrum, so the 5008.24 A line alone defined the redshift for this region. As mentioned in

Section 3 the location of longslit in the LUCIFER setup was optimized for detection of R2

and C15, so it is not surprising that we did not have a detection for C11 in the LUCIFER

data. All these [O iii] redshifts were corrected for the earth’s motion using topocentric radial

velocities4 appropriate for the date and location of the observations. The error bars on the

redshift are a compilation of the RMS from wavelength calibration during data reduction,

the 1 sigma error on the best-fit central wavelength from Gaussian fitting, averaging of two

redshifts when applicable, and a 0.02 km s−1 uncertainty on the topocentric radial velocities.

See Table 1 for a summary of this data.

5.2. Lyα Redshifts

To determine the redshift of the Lyα emission line, we followed the same method-

ology we previously used in McLinden et al. (2011). Namely, the Lyα profiles for C11

and C15 were fit with a single asymmetric Gaussian plus constant, using IDL routine

ARM ASYMGAUSSFIT5. These fits are overlaid on the extracted 1D Lyα spectra in Figure

5. The asymmetric Gaussian fitting routine allows for, but does not require, that the fit be

asymmetric. The central wavelength of the best-fit asymmetric Gaussian defines the Lyα

redshift.

The error on the line center was determined from 1000 Monte Carlo iterations. In each

iteration the flux at each wavelength was altered by a random amount proportional to the

error bar at that point and the altered spectrum was re-fit with an asymmetric Gaussian.

The standard deviation of the 1000 best-fit line centers is used as the 1σ error on the Lyα

line center.

We note that the SAURON IFU datacube comes from a combination of data obtained

4http://fuse.pha.jhu.edu/support/tools/vlsr.html

5http://hubble.as.arizona.edu/idl/arm/arm asymgaussfit.pro

– 12 –

in 2002 (Bower et al. 2004) and 2006 (Weijmans et al. 2010). The data were not corrected

to the local standard of rest (LSR) before combination into a single datacube. Therefore,

the Lyα profiles derived from the combined data may be slightly broadened with a slightly

shifted center compared to the intrinsic Lyα profile. To account for this effect, we calculated

the corrections that should be applied to the 2002 (∼ -26.6 km s−1) and the 2006 data

(∼ +2.1 km s−1)6. Combining two such Gaussians, offset from one another by ∼ 28.7 km

s−1, means that the intrinsic line center (if they had both been calibrated to the LSR before

combination) is shifted by ∼ 12.3 km s−1. We therefore fold this 12.3 km s−1 term into

our error calculations for redshift of the Lyα line and the subsequently derived [O iii]-Lyα

velocity offsets. Table 1 summarizes our Lyα and [O iii] redshift information.

Fig. 5.— Best fit Gaussians for C15 (top row) and C11 (bottom row) from

ARM ASYMGAUSSFIT show in blue. Lyα spectra are shown in black. Bottom right

panel shows C11 simultaneously fit with a double Gaussian using MPFITEXPR - the two

Gaussians are shown in blue and red, and their sum is shown in orange.

6http://fuse.pha.jhu.edu/support/tools/vlsr.html

– 13 –

5.3. Velocity Offsets

We derive a velocity offset (∆v) between the [O iii] and Lyα lines by comparing the

redshifts for each line as derived from the central wavelength of the best-fit Gaussians to the

two lines, where the fitting procedure is described above. We consider the [O iii] redshift

to be the systemic redshift. This method finds an offset between [O iii] and Lyα in C15 of

-72.4 ± 41.6 km s−1 from the LUCIFER data and -51.3 ± 42.1 km s−1 from the NIRSPEC

data. These results are remarkably consistent with one another, in spite of being measured

in slightly different apertures with different instruments. These results are modestly negative

though also very nearly consistent with zero. We find in ∆v is 5.8 ± 32.9 km s−1 for C11

in the NIRSPEC data, which is consistent with no offset. See Table 1 for a compilation of

these results.

One may argue that the rather broad Lyα line in C11 may be better fit with a double

Gaussian profile, especially if one considers that the bump to the left of the highest peak is

not noise, but in fact evidence of a second, unresolved peak. This could be a blue bump that

is not fully resolved and separated from the main red peak. Or it could be the smaller red

peak at v=0, as in Fig. 12 in Verhamme et al. (2006). To consider these possibilities, we fit

C11 a second time, simultaneously fitting two Gaussians plus a constant (see bottom right

panel of Figure 5). When we do this we find that the right peak yields only a modest offset,

where the Lyα line is offset by ∼ 18 km s−1 from the [O iii] redshift of C11 from NIRSPEC.

The bluer Lyα peak is blueshifted with respect to [O iii] by ∼ 808 km s−1. This leads to a

rather inexplicably large offset between the two Lyα peaks, ∼ 826 km s−1, especially when

the red peak is so mildly offset, which may disfavor this secondary interpretation of the

modest bump as a blue bump. While acknowledgement of the additional fit described above

is worthwhile, the discussion stays much the same, the observed velocity offset between [O iii]

(systemic) and Lyα is at best modestly negative and/or may be consistent with zero.

5.4. [O iii] Flux in C15 and R2

As previously described in Section 5.1, the [O iii] lines are fit with a symmetric Gaussian

plus a constant. The resulting area under the Gaussian gives us a line flux measurement for

each line. We find an [O iii] flux of 1.1 ± 0.1 × 10−16 erg s−1 cm−2 for C15 from LUCIFER

(5008 A line only) . For C15 in NIRSPEC we find line fluxes of 1.6 ± 0.5 × 10−17 erg s−1

cm−2(4960 A) and 4.6 ± 0.5 × 10−17 erg s−1 cm−2(5008 A). Finally, for C11 from NIRSPEC

we measure a line flux of 1.6 ± 0.4 × 10−17 (5008 A line only).

Since we do not detect an [O iii] line in R2 we instead measure an [O iii] line flux upper

– 14 –

limit of ≤ 3.4 × 10−17 erg s−1 cm−2. LUCIFER has sufficient wavelength coverage to place

upper limits on both [O ii] and Hβ in our C15 spectrum. We derive upper limits for [O ii]

and Hβ in C15 of 1.3 × 10−17 and 2.2 × 10−16 erg s−1 cm−2 from our LUCIFER spectrum.

This combination of a measured [O iii] line flux and upper limits on [O ii], and Hβ line flux

in C15 allow us to put constraints on the metallicity of the LBG embedded in C15 using

R23, where R23 is defined as[O ii]+[O iii]

Hβ(Pagel et al. 1979; Kewley & Dopita 2002). For

C15, R23 > 0.76 given the line flux measurements and limits described above. NIRSPEC

wavelength coverage only allows us to place upper limits on Hβ. We find Hβ upper limits

of 3.7 and 7.2 × 10−18 erg s−1 cm−2 for C15 and C11, respectively.

To compute the 3σ line flux upper limit for R2 quoted above, we added a mock Gaussian

emission line to the spectra to represent [O iii], similar to the procedure in Finkelstein et al.

(2011b). The sigma of the Gaussian was fixed to 5.52 A, or the σ from our faintest [O iii]

detection to date (McLinden et al. 2011). The area under the mock line was measured

using a symmetric Gaussian, this area determines the line flux of the mock line. Then we

determined the noise on the line flux measurement from 1000 Monte Carlo iterations, where

the flux array was modified each time by a random amount proportional to the error bars.

We repeated this process, each time decreasing the area under the mock Gaussian until the

signal to noise ratio (SNR) dropped below 5σ. The line flux in the mock line where the SNR

crossed below 5σ determines σ. However, because one cannot know, without an nebular

emission line measurement for reference, exactly how much, if any the Lyα line is offset from

the [O iii] line, we repeat this calculation, fixing the mock line at different redshifts to mimic

different velocity offsets. We found the 3σ [O iii] line flux detection limit as an average of

this technique from 6 different redshifts corresponding to velocity offsets of 0-500 km s−1, in

increments of 100 km s−1. A range of 0 - 500 km s−1 was chosen to mirror the magnitude

of Lyα - [O iii] velocity offsets we have observed of 52 - 342 km s−1 in three z ∼ 3.1 LAEs.

Table 1. Comparison of [O iii] and Lyα redshifts

[O iii] Lyα

Region 4960.295 5008.24 zavg zcorr z ∆v (km s−1)

C15 (LUCIFER) 3.0983 3.0986 ± 0.0003494 3.0976 ± 0.0004159 -72.4 ± 41.6

C15 (NIRSPEC) 3.0973 3.0980 3.0976 3.0974 ± 0.0005095 3.0967 ± 0.0002070 -51.3 ± 42.1

C11 (NIRSPEC) 3.1002 3.0999 ± 0.0003590 3.1000 ± 0.0002138 5.8 ± 32.9

Note. — Comparison of [O iii] and Lyα redshifts in C11 and C15 as measured with LUCIFER and NIRSPEC. zavg is listed

when the redshifts from 4960.295 and 5008.24 A lines were averaged. zcorr is the redshift after correction to the LSR. ∆v is

velocity offset between [O iii] and Lyα.

– 15 –

The range of 3σ detection limits over this wavelength range was 3.12 - 3.54 × 10−17 erg s−1

cm−2. The [O ii] and Hβ upper limits described above are found using the same procedure,

except the upper limits need only be derived at a single fixed wavelength for each line, the

wavelength defined by the [O iii] redshift.

Given [O iii] line flux detections in C15 and C11 and an [O iii] upper limit in R2, we can

compare the nebular emission from these subregions. This is of interest because C15 and

C11 are associated with LBGs embedded within the larger LAB1 halo structure, whereas

R2, in spite of its stronger Lyα emission , is not associated with any underlying galaxy

(Weijmans et al. 2010). The ratio of Lyα luminosity in C15 to R2 is 0.7 (Weijmans et al.

2010). We find that the ratio of [O iii] in C15 (LUCIFER) to R2 is ≥ 3.4 and the ratio of

[O iii] in C15 (NIRSPEC) to R2 is ≥ 1.4, meaning that while R2 is brighter in Lyα C15 is

brighter when looking at [O iii] nebular emission. Finally, the [O iii] to Lyα ratio measured

in the region without an LBG is therefore at least 1.9 – 4.5 times smaller than the same

ratio measured in the LBG. This measurement would likely indicate that something other

than star formation is powering the Lyα emission in region R2 and that there may very well

be different sources powering Lyα emission in different regions of the same blob.

C11 presents a somewhat different story. The ratio of Lyα luminosity in C11 to R2 is

1.0 (Weijmans et al. 2010). Then the ratio of [O iii] in C11 (from NIRSPEC) to R2 is only

≥ 0.5. The [O iii] to Lyα in C11 is only 0.06, which is smaller than that same measurement

in R2, where the ratio is ≤ 0.13. These divergent results are interesting, perhaps giving

further credence to the idea that there are different processes at work in different regions of

this blob.

5.5. Asymmetry and [O iii] - Lyα Offset

We have now measured the velocity offset between [O iii] and Lyα in 3 z ∼ 3.1 LAEs

(McLinden et al. 2011) and added 3 ∆v measurements for LAB1 in this work. Amongst

these measurements, the regions C11 and C15 in LAB1 have the smallest absolute ∆v mea-

surements and are the only negative values. This is in contrast to three z ∼ 3.1 LAEs in

which we found offsets ranging from 52 – 342 km s−1. Given this information, we can com-

pare another signature of outflows, namely asymmetry in the Lyα profile, with the velocity

offset measurements.

We quantify asymmetry as σred/σblue where σred is the sigma of the red side of the

best-fit asymmetric Gaussian and σblue is the sigma on the blue side of the best-fit asymmet-

ric Gaussian, where both are parameters returned by the ARM ASYMGAUSSFIT routine.

– 16 –

Fig. 6.— Plot of Lyα (blue) over [O iii] (black) where Lyα has been shifted to [O iii] frame

via Lyα[O iii] = Lyαobserved × 5008.241215.67

. Observed spectra are shown with solid lines, best

fit Gaussians for both Lyα and [O iii] are shown with dashed lines. Top panel is C15, left

plot shows [O iii] from LUCIFER, right plot shows [O iii] from NIRSPEC. Bottom panel is

C11 from NIRSPEC.

– 17 –

With this definition, a profile with asymmetry > 1.0 is considered asymmetric, and the

asymmetry is dominated by the red-wing. Objects with asymmetry = 1.0 are symmet-

ric, ≤ 1.0 have blue-wing dominated asymmetry. The red-wing dominated asymmetry is

the expected direction of the asymmetry in the Lyα line from high-z galaxies, as the red

side of the line can be enhanced in the presence of an expanding shell (Verhamme et al.

2006; Dijkstra & Wyithe 2010) and/or by interaction with neutral Hydrogen in the IGM

(Rhoads et al. 2003; Dawson et al. 2004). Measured in this way C11 has an asymmetry

of 0.74 ± 0.02 and C15 has an asymmetry of 0.9 ± 0.03 and 1.1 ± 0.03 (LUCIFER and

NIRSPEC apertures, respectively). An asymmetry of < 1.0, like C11, can suggest infalling

material. However, the Lyα-[O iii] velocity offset of 5.8 ± 32.9 km/s in C11 is not very

supportive of such an interpretation. In three z ∼ 3.1 LAEs in which we have measured a

velocity offset between [O iii] and Lyα we find asymmetries of 0.97 ± 0.1, 1.04 ± 0.1, and

1.65 ± 0.1 for LAE7745, LAE27878 and LAE40844, respectively (McLinden et al. 2011). We

add four additional asymmetry data points by using four z ∼ 2 LAEs from Hashimoto et al.

(2012). Hashimoto et al. (2012) measured redshifted Lyα lines in these objects with respect

to Hα lines in the same objects. They report velocity offsets of 18 – 190 km s−1 in their

four LAEs, in good agreement with our range of 52 – 342 km s−1 in LAEs. We measured

the asymmetry of the Lyα lines presented by Hashimoto et al. (2012) in the same man-

ner as above, by fitting each line with asymmetric Gaussian and quantifying asymmetry as

σred/σblue. Measured in this way the LAEs from Hashimoto et al. (2012) have asymmetries

of 0.53 - 1.7.

Figure 7 demonstrates a possible trend where asymmetry in the Lyα profile increases

with increasing velocity offset. We find a moderate Pearson linear correlation coefficient

of 0.427 (P=0.110) between the velocity and asymmetry values, suggesting the trend may

be real. This velocity-asymmetry correlation is not unexpected, as increasing (red-wing

dominated) asymmetry in the Lyα profile is tied to increasing shell expansion velocities in

Verhamme et al. (2006), when outflows are modeled with a central monochromatic source

and an expanding shell. This is because Lyα photons are seen as more redshifted by the

neutral hydrogen in the shell with higher shell expansion speeds, decreasing the cross-section

for interaction (Verhamme et al. 2006). A symmetric line, as seen in C15, is not expected in

models with large outflows, but could be consistent with static/nearly-static profiles - if the

two symmetric peaks produced from a static slab or shell scenario (Verhamme et al. 2006)

are unresolved. The authors note that the point attributed to LAE40844 strongly influences

this correlation. Removal of this point and recalculation of the Pearson coefficient with the

remaining eight points only yields a coefficient of 0.170 (P=0.331), which does not suggest

any significant correlation.

– 18 –

6. DISCUSSION

6.1. Comparison with Blobs at z ∼ 2

We have measured ∆v between [O iii] and Lyα in two subregions of LAB1 and found

that ∆v is modestly negative or consistent with 0 km s−1. These measurements seem to

downplay the role of powerful outflows in powering the Lyα in these regions and could even

hint at some infall. Interestingly, this measurement of ∆v = small is not the first time this

phenomenon has been reported in a Lyα blob, suggesting this is an important phenomenon

that must be explored to better understand the nature of high-z LABs. Yang et al. (2011)

investigated two z ∼ 2.3 LABs, where one blob (called CDFS-LAB02) had ∆v ∼ 230 km s−1

and the other blob had ∆v consistent with zero (called CDFS-LAB01). The velocity offsets

reported in Yang et al. (2011) were measured by comparing the redshift of Hα to that of

Lyα, a very similar tactic to our comparison of [O iii] to Lyα, as both [O iii] and Hα are

nebular emission lines and therefore probe the same regions.

6.2. Comparison to LAEs and LBGs

Our ∆v measurements, and those of Yang et al. (2011) , are significantly less than the

larger velocity offsets typically seen in Lyman break galaxies (LBGs) and even Lyα emitting

galaxies (LAEs), velocity offsets which are typically interpreted as clear signatures of strong

winds in these galaxies. Steidel et al. (2010) report median velocity offsets between Hα and

Lyα of 445 km s−1 in 41 z ∼ 2.3 LBGs. And even LAEs, whose typical velocity offsets have

been found to be smaller, have ∆v as large as 342 km s−1(McLinden et al. 2011).

Steidel et al. (2010) also report a median velocity offset between Hα and strong inter-

stellar absorption lines of -164 km s−1 in 86 LBGs, which further supports interpretations

of the presence of outflows in LBGs, since the blue-shifted absorption implies absorption in

material moving towards the observer. Shapley et al. (2006) previously measured a redshift

from low-ionization interstellar absorption (LIS) lines in the LBG C11, finding that the ab-

sorption lines are offset from the Lyα line by -380 km s−1. Using the LIS redshift of 3.0962

from Shapley et al. (2006) and comparing this to the Lyα redshift we derive for C11 yields

an offset of ∼ 278 km s−1. Comparing our [O iii] (systemic) redshift for C11 (3.0999) to

the LIS redshift, we can estimate that the LIS lines in C11 are offset from [O iii] by ∼ -270

km s−1. This comparison of Lyα and [O iii] redshifts to LIS absorption redshifts provides a

stronger signature of an outflow than we get when comparing Lyα and [O iii]. In fact, it is

particularly interesting that the magnitude of this second signature of winds (i.e. blueshifted

interstellar absorption lines) is so similar when comparing the Steidel LBGs to C11, yet the

– 19 –

magnitude of the offset between nebular emission lines (Hα or [O iii]) and Lyα offset are so

different. We note that the lack of ∆v between [O iii] and Lyα does not have to rule out

some outflows in LAB1. Rather, the lack of a Lyα-[O iii] offset may just imply that outflows

are not a significant mechanism for helping Lyα photons escape.

6.3. Previous Studies of LAB1

This phenomena, i.e. ∆v = small and/or = 0 km s−1, leads to the question of whether

the lack of substantial velocity offset between Hα-Lyα or [O iii]-Lyα in these blobs in fact

rules out outflows or if there is some, yet to be understood phenomena, that damps or erases

this particular wind signature. This question is particularly relevant given the recent report

from Hayes et al. (2011) that there is polarized radiation emanating from LAB1, polarization

that is indicative of scattering of Lyα photons at large radii with respect to their site of

production. This may be a sign of outflows helping to drive the Lyα photons to these large

radii, but we are not seeing the velocity offsets between nebular emission lines and Lyα that

we would expect to see if this was the case, velocity offsets we have been able to see in other

objects at similar redshifts believed to have strong winds. In addition, Bower et al. (2004)

and Weijmans et al. (2010) both measure a velocity shear in the Lyα emission from C11 and

C15. As the authors point out, such a velocity shear could be consistent with infalling gas,

outflowing gas and/or rotation of the system, and such scenarios cannot be differentiated

from the Lyα data alone. While both papers use this velocity shear to argue in favor of

the presence of outflows in C11 and C15, we can report no signature of such outflows when

we compare the redshifts of [O iii] and Lyα, a comparison that has proven to be a useful

diagnostic of winds in LAEs and LBGs at similar redshifts.

6.4. Comparison to Radiative Transfer Models

We explored available Lyα radiative transfer models to see if there were any models that

might shed light on the physical conditions that could lead to a modestly negative ∆v or ∆v

of ∼ 0 km s−1 between [O iii] (or Hα) and Lyα. We focused on the Lyα profiles produced by

Verhamme et al. (2006) (henceforth V06) from their 3D Monte Carlo Lyα radiative transfer

code. V06 explore a variety of physical conditions and geometric orientations to explore

the diversity of Lyα profiles that arise from different conditions. See V06 for extensive

details on these models and the model parameters. We find, however, that none of the

models presented in V06 are in great agreement with our observations (or the observations

of Yang et al. (2011)). The only models that are marginally consistent with our observations

– 20 –

are those that have two significant Lyα peaks, where the centroid of those two peaks is

centered at v ∼ 0, with one peak redward of v= 0, and the other blueward. Convolved

with an appropriate Gaussian to mimic the instrumental resolution of SAURON (290 km

s−1), these models somewhat resemble the Lyα profiles presented here. This double-peaked,

centered at v ∼ 0 profile occurs when a central monochromatic source is embedded in a

static slab (with or without dust, V06 Figure 3), a central monochromatic source sits in a

non-expanding shell (V06 Figure 14), and when a single monochromatic source sits in an

expanding shell with very small velocity gradient (Vmax = 20 km s−1, V06 Figure 7). The

closest match (see Figure 8) is from V06, Figure 14 with a static shell. In addition, we

explored whether the infalling halo models of V06 might agree with our observations, these

models are presented in Figure 9. There may be some qualitative agreement between the

models and observed profiles here, but the models explored are too far blueward of v=0 to

match the Lyα profiles we present here. In either scenario (expanding or infalling) the main

peak is expected to be asymmetric, which is inconsistent with the Lyα profiles of at least

C15 presented here. We contend that more modeling of Lyα radiative transfer with direct

applications to the observations we have presented here and those observations presented in

Yang et al. (2011) needs to be done, to better understand the physical conditions, geometry,

and kinematics that can produce single peaked Lyα lines, with ∆v values that are modestly

negative or consistent with zero.

7. CONCLUSION

We have measured [O iii] in two subregions of LAB1, C11 and C15, regions that are

associated with underlying LBGs within the larger halo structure. We have quantified the

velocity offset between [O iii] and Lyα redshifts in these regions, finding that both mea-

surements are modestly negative or consistent with zero. This is an intriguing result since

powerful outflows have been proposed as possible ways to explain the luminosity and large

spatial extent of LABs. We cannot completely rule out the presence of strong winds and

outflows in LAB1, but we can state that we do not see two typical markers of their presence.

(1) The aforementioned result that we do not find that Lyα is redshifted with respect to

[O iii] in the two LAB1 subregions C15 and C11. And (2) we do not measure strong red-wing

dominated asymmetries in the Lyα profiles of these objects, in contrast with z ∼ 2 and z ∼

3.1 LAEs where the asymmetry of the Lyα line appears to increase with increasing velocity

offset between [O iii] (or Hα) and Lyα. If outflows are present in LAB1, they do not appear

to be a crucial mechanism driving Lyα escape.

In addition to the conclusion above, we have placed an upper limit on [O iii] line flux

– 21 –

from region R2, a subregion of LAB1 unassociated with any known galaxy and compared

this to the [O iii] flux from subregion C15, which is associated with an LBG. We find that in

spite of the strong Lyα emission present in R2, the [O iii] flux from C15 is stronger than that

of R2 by at least 1.4 – 2.5 times. This measurement may indicate that diverse sources of

Lyα emission may be responsible for powering different regions even within the same blob.

The LBT is an international collaboration among institutions in the United States, Italy

and Germany. LBT Corporation partners are: The University of Arizona on behalf of the

Arizona university system; Instituto Nazionale di Astrofisica, Italy; LBT Beteiligungsge-

sellschaft, Germany, representing the Max-Planck Society, the Astrophysical Institute Pots-

dam, and Heidelberg University; The Ohio State University, and The Research Corporation,

on behalf of The University of Notre Dame, University of Minnesota and University of Vir-

ginia.

This work was supported by a NASA Keck PI Data Award (award number is 1408709),

administered by the NASA Exoplanet Science Institute. Data presented herein were obtained

at the W. M. Keck Observatory from telescope time allocated to the National Aeronautics

and Space Administration through the agency’s scientific partnership with the California

Institute of Technology and the University of California. The Observatory was made possible

by the generous financial support of the W. M. Keck Foundation. The authors wish to

recognize and acknowledge the very significant cultural role and reverence that the summit

of Mauna Kea has always had within the indigenous Hawaiian community. We are most

fortunate to have the opportunity to conduct observations from this mountain.

NSF grant AST-0808165 also provided funding for this work.

Thanks to Mark Richardson for various helpful conversations. Thanks also to Richard

Bower for his contributions to the SAURON data and for sharing this data.

REFERENCES

Ageorges, N., Seifert, W., Jutte, M., et al. 2010, “LUCIFER1 commissioning at the LBT”,

SPIE Proc. Vol. 7735-56

Bower R. G., Morris, S. L., Bacon, R., et al., 2004, MNRAS, 351, 63

Chapman S. C., Lewis G. F., Scott D., et al. 2001, 548, 17

Chapman, S. C.,Windhorst, R., Odewahn, S., et al., 2003 ApJ, 599, 92

– 22 –

Chapman S. C., Scott D., Windhorst R. A., et al. 2004, ApJ, 606, 85

Cowie, L. L., & Hu, E. M. 1998, AJ, 115, 1319

Dawson, S., Rhoads, J. E., Malhotra, S., et al. 2004, ApJ, 617, 707

Dey, A., Bian, C., Soifer, B. T., et al. 2005, ApJ, 629, 654

de Zeeuw P. T., Bureau, M., Emsellem, E., et al., 2002, MNRAS, 329, 513

Dijkstra M. & Loeb A., 2009, MNRAS, 400, 110

Dijkstra, M. & Wyithe, J. S. B. 2010 MNRAS, 408, 352

Finkelstein, S. L., Cohen, S. H., Moustakas, J., et al. 2011 ApJ, 733, 117

Francis, P. J., Williger, G. M., Collins, N. R., et al. 2001, ApJ, 554, 1001

Geach J. E., Smail, I., Chapman, S. C., et al. 2007, ApJ, 655L, 9

Geach J. E., Alexander, D. M., Lehmer, B. D., et al. 2009, ApJ, 700, 1

Haiman Z. & Spaans M., Quataert E., 2000, ApJ, 537, 5

Hashimoto, T., Ouchi, M., Shimasaku, K., et al. 2012, submitted to ApJ

Hayes, M., Scarlata, C. & Siana, B. 2011, Nature, 476, 304

Hu, E. M., Cowie, L. L., & McMahon, R. G. 1998, ApJ, 502, L99

Kelson, D., 2003, PASP, 115, 688

Kewley, L. J. & Dopita, M. A. 2002, ApJS, 142, 35

Kohno, K., Tamura, Y., Hatsukade, B., et al. 2008, ASPC, 399, 264

Matsuda Y., Yamada, T., Hayashino, T., et al. 2004, AJ, 128, 569

McLean, I. S., Becklin, E. E., Bendiksen, O., et al. SPIE, 1998, 3354, 566

McLinden E., Finkelstein, S., Rhoads, J., et al. 2011, ApJ, 730, 136

Nilsson, K. K.; Fynbo, J. P. U.; Mller, P.; Sommer-Larsen, J.; Ledoux, C. 2006, A&A, 452,

L23

Ohyama Y., Taniguchi, Y., Kawabat, K. S., et al. 2003, ApJ, 591, 9

– 23 –

Pagel, B. E. J., Edmunds, M. G., Blackwell, D. E., et al. 1979, MNRAS, 189, 95

Pickles, A. J. 1998, PASP, 110, 863

Prescott, M. K., Smith, P. S., Schmidt, G. D. & Dey, A. 2011, ApJ, 730, 25

Prescott, M. K., Dey, A., Jannuzi, B. T., 2012, ApJ, 748, 125

Rhoads, J. E., Malhotra, S., Dey, A., et al, ApJ, 545L, 85.

Rhoads, J. E., Dey, A., Malhotra, S., et al. 2003, AJ, 125, 1006

Rousselot, P., Lidman, C., Cuby, J.-G., et al. 2000, A&A, 354, 1134

Seifert, W., Appenzeller, I., Baumeister, H., et al. 2003, “LUCIFER: a Multi-Mode NIR

Instrument for LBT”, SPIE Proc. Vol. 4841 Instrument Design and Performance for

Optical/Infrared Ground-Based Telescopes

Shapley, A. E., Steidel, C. C., Pettini, M., et al. 2006, ApJ, 651, 688

Spergel, D. N., Bean, R., Dore, O., et al. 2007, ApJS, 170, 377

Steidel, C. C., Adelberger, K. L., Dickinson, M., et al. 1998, 492, 428

Steidel, C. C., Adelberger, K. L., Shapley, A. E., et al. 2000, ApJ, 532, 170

Steidel, C. C., Erb, D. K., Shapley, A. E., et al. 2010, ApJ, 717, 289

Taniguchi Y. & Shioya Y., 2000, ApJ, 532, 13

Verhamme, A., Schaerer, D. & Maselli, A. 2006, A&A, 460, 397

Weijmans, A-M., Bower, R. G., Geach, J. E., et al. 2010, MNRAS, 402, 2245

Yang, Y., Zabludoff, A., Jahnke, K., et al. 2011, ApJ, 735, 87

Yang, Y., Decarli, R., Dannerbaurer, H., et al. 2012, ApJ, 744, 178

This preprint was prepared with the AAS LATEX macros v5.2.

– 24 –

Fig. 7.— Asymmetry of the Lyα profile as a function of velocity offset in five Lyα-emitting

objects at z ∼ 3.1 (shown in black) and four z ∼ 2 objects (shown in red). Three LAEs

from McLinden et al. (2011) are labeled with the prefix ‘LAE.’ Red points with the prefix

‘cosmos’ or ‘cdfs’ are from Hashimoto et al. (2012). Overall, asymmetry may increase with

increasing velocity offset.

Fig. 8.— Lyα profiles (black dashed line) compared to V06 outflowing shell and halo models

(solid blue). V06 models have been convolved with a Gaussian to match instrumental resolu-

tion of SAURON. Top row shows comparison to model in V06 Figure 14. Bottom row shows

comparison to model in V06 Figure 7. While these models and those in Figure 9 represent

the closest matches to the observed profiles, clearly none are an excellent fit.

– 25 –

Fig. 9.— Lyα profiles (black dashed line) compared to V06 infalling halo models (solid

blue). V06 models have been convolved with a Gaussian to match instrumental resolution

of SAURON. Top row shows comparison to model in left panel of Figure 5 in V06, bottom

row shows comparison to model in right panel of Figure 5. While these models and those in

Figure 8 represent the closest matches to the observed profiles, clearly none are an excellent

fit.